Futarchy & Decision markets for deciding next Devcon

Devcon Improvement Proposals (DIPs)
1

Summary of Proposal

Use prediction market forecasts to help decide the location of Devcon 2026. Launch prediction markets about the location for 2026, and offer conditional markets for each potential location forecasting the attendance, attendee survey responses or other information that would be helpful to event organizers.

Abstract

Vitalik has written about using prediction markets for futarchy and info finance, but with “conditional markets” we can now implement them for the types of use cases he envisions. Seer One will offer a prediction market about the eventual location of Devcon 2026. It will be a categorical market allowing traders to predict one correct answer from multiple possible locations. Each potential outcome (i.e., location for Devcon 2026) will have conditional markets attached to it. Traders will then buy and sell shares for metrics (i.e., attendance and satisfaction) conditional on each potential location.

For example, if the only two potential locations for 2026 were Durban and Buenos Aires, then the parent market would be:

  • What will be the location of Devcon 2026?
    • Durban
    • Buenos Aires

One pair of conditional markets would be “scalar markets” forecasting attendance. These resolve to an integer on a set scale:

  • If Durban: How many people will attend Devcon 2026? [Scalar, 0-10,000]
  • If Buenos Aires: How many people will attend Devcon 2026? [Scalar, 0-10,000]

Here is an example of a 2024 Devcon survey question:

Another pair of conditional markets would predict attendee satisfaction based on their survey response:

  • If Durban: What will be the average “overall experience” rating based on Devcon survey responses? [Scalar, 1-5]
  • If Buenos Aires: What will be the average “overall experience” rating based on Devcon survey responses? [Scalar, 1-5]

Organizers can compare the attendance forecasts for each city and incorporate that information in their decision-making for Devcon. For example, the conditional markets could forecast:

Potential Location Expected Attendance Expected Satisfaction
Durban 5,652 4.22
Buenos Aires 7,119 4.33

Hypothetically, Durban and Buenos Aires both predict similar Overall Experience (“Satisfaction”) ratings from attendees surveys, with Buenos Aires receiving a slightly higher score. However, the predicted attendance for Buenos Aires is significantly higher than Durban. Organizers can take these forecasts into account with the other information being used for making a decision about the location for Devcon.

Motivation & Rationale

Devcon strives for decentralization and now can crowd-source forecasts about its own events. Prediction and conditional markets are powerful tools that leverage collective wisdom to estimate the impact of decisions. Forecasts utilizing the ‘Wisdom of the Crowd’ have been consistently proven more accurate than experts in countless fields including economics, politics, public health, meteorology, and more.

Prediction markets will give the Ethereum community a voice in deciding the next location, raise awareness about the event and engage ethereum developers around the world. The forecasts generated by the conditional markets will provide valuable information about the potential destinations under consideration.

Implementation

The Seer prediction marketplace is on both Ethereum and Gnosis chain and utilizes the Gnosis Conditional Token Framework. The Devcon 2026 markets could be launched within 1-2 weeks and no funding is requested.

Operational Requirements & Ownership

Market implementation and operation will be provided by Seer One. Market resolution will be by Reality.eth and arbitration through Kleros Court. The only requirements would be:

  • Make public the potential cities for Devcon 2026 to allow the creation of the prediction market.
  • Have decision makers responsible for the location choice look at the futarchic estimates before making their decisions so that they can use those as one of the elements guiding their decisions.
  • Reveal the date of the announcement at least a couple of days before the actual announcement.
  • Commit to report the attendance and “overall experience” survey question data within a reasonable time after the event.

Links

7 Likes
2

This is not a requirement, but it would be great to have the historical average and variance of the “overall experience” answer from previous post-devcon surveys. The market can work without it but:

  • If the information is not public, forecasters may spend a significant share of their time in estimating the average (like running their own polls).
  • If satisfaction scores are always within a particular range (like 3-5), we could restrict the market to this range to be more efficient.
1 Like
3

We are very much open to doing a Futarchy and Decision markets experiments via DIP(s) for DEVCon(nect). That said I do not believe the location is a good use case. This can actually be counter-productive for us as it puts us in a worse position making deals with venues and hotels. Also not sure it would have desireable outcomes - e.g. I think prediction markets would have signaled against e.g. going to Bogota - but I think it was important and good we did so.
I think we should explore other options to use Futarchy and Decision markets to improve DEVCon.

1 Like
4

For this proposal the Devcon team just needs to look at the results as one source of information among others, not necessarily follow them, so the risk is very limited.
Also keep in mind that it’s a market, not a vote, so if most people believed Bogota was gonna be a bad location but some believed, with high conviction, that it would be a good one, they could influence the market way more than the masses.
If you believe the market is wrong, you can “correct it” (and profit from it if you end up right).

This is true for markets about the location (if venues and hotels know a location is “locked in”, they could offer worse deals), but those will exist independently of the Devcon team looking a futarchy markets (for example this market which was created due to enthousiastic Argentinian community members independantly of this proposal Seer | Will the Devconnect 2025 conference be hosted in Greater Buenos Aires, Argentina?).
Having markets on the rating given conditionally on the location is very unlikely to have an impact on venue pricing (they could in theory signal that a location with a higher expected rating is more likely to chosen, but that seems more of a theoretical thing (I don’t expect much people outside of crypto/decision specialists to be doing those inferences)).

(for context, we talked on X about futarchy for Devcon and ligi was proposing attendance markets)

I do think attendance markets would be nice. This can allow assigning rooms depending of the expected attendance. Those would be simpler (we just need an attendance market for each talk, no complex conditional markets) which can be better for first time PM participants.
However, this would be a bigger commitment for the organising team as we would need to get a report of attend1nce. However this is possible is possible (for example, just by pointing the camera on the audience, participants can be counted after the event and Seer One could provide some workers to count participants from the recordings).

We initially proposed rating based on location markets as it required minimal efforts from the team (just indicating potential locations and reporting the rating of the feedback form).
If you are up to, we could switch to attendance markets (or if you have other usecases, we can look at those too), but we need to be sure attendance will be available.

2 Likes
5

I think we should switch to attendance markets as I still see more upside and less downside for DEVCon here in contrast to markets around the location. Also this could lead to more iterations as there can be more markets than just one / event.
We just need to think about the reporting part. Ideally we find a more privacy preserving way - other than pointing a camera on attendees (even though we are thinking about cameras in the stages streaming to our mission-control-room to get a better overview on what is going on in the venue - but this already feels a bit wrong - and publishing the content (which would be needed here as far as I understand) - feels even worse).